Systems and methods for concept based searching or recommendation are disclosed. More particularly, embodiments of a concept based approach to the search and analysis of data, including the creation, update or use of concept networks in searching and analyzing data are disclosed, including embodiments of the usage of such concept networks in artificial intelligence systems that are capable of utilizing concepts expressed by users to return or evaluate associated images.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The system of claim 1, wherein determining the destination associated with the canonical concept is determined comprises: determining a score for the canonical concept for each destination in the set of destinations, wherein determining a score for the canonical concept for the destination comprises accessing the destination concept vector associated with the destination to determine the score for the canonical concept for the destination and selecting the destination based on the score for the canonical concept for each destination in the set of destinations.
3. The system of claim 1, wherein the destination is a neighborhood, apartment, or house.
4. The system of claim 1, wherein the current interaction is a natural language interaction.
5. The system of claim 1, wherein the image concept vector for the image was determined by: determining one or more concepts associated with the image; determining a score for the one or more concepts associated with the image and applying the score for the one or more concepts associated with the image to a concept net relationship matrix.
6. The system of claim 1, wherein the canonical concept is based on one or more previous user interactions in a session associated with the user.
The system relates to personalized user interaction management, specifically improving user experience by dynamically adapting to individual behavior patterns. The core problem addressed is the lack of contextual awareness in traditional systems, which often fail to account for a user's evolving preferences and actions within a session. This leads to inefficiencies, such as irrelevant recommendations or repetitive prompts, degrading usability. The system leverages a canonical concept—a standardized representation of user intent or context—derived from prior interactions within the current session. By analyzing these interactions, the system identifies recurring patterns or preferences, enabling real-time adjustments to system responses. For example, if a user frequently accesses specific features or inputs similar queries, the system prioritizes these elements to streamline future interactions. This approach ensures that the system remains aligned with the user's immediate needs, reducing friction and enhancing engagement. The canonical concept is dynamically updated as new interactions occur, allowing the system to adapt continuously. This contrasts with static profiles or historical data, which may not reflect the user's current context. The system may also integrate additional data sources, such as device usage or environmental factors, to refine its understanding of the user's intent. The result is a more intuitive and responsive interaction framework, particularly valuable in applications like virtual assistants, recommendation engines, or adaptive interfaces.
7. The system of claim 1, wherein the image is selected based on its similarity to other images in the set of images.
9. The non-transitory computer readable medium of claim 8, wherein determining the destination associated with the canonical concept is determined comprises: determining a score for the canonical concept for each destination in the set of destinations, wherein determining a score for the canonical concept for the destination comprises accessing the destination concept vector associated with the destination to determine the score for the canonical concept for the destination and selecting the destination based on the score for the canonical concept for each destination in the set of destinations.
10. The non-transitory computer readable medium of claim 8, wherein the destination is a neighborhood, apartment, or house.
A system and method for optimizing delivery routes involves determining an optimal path for a delivery vehicle to reach a destination, which may be a neighborhood, apartment, or house. The system analyzes real-time traffic data, road conditions, and delivery constraints to generate an efficient route. It also considers factors such as delivery time windows, vehicle capacity, and customer preferences to improve delivery efficiency. The system may use machine learning algorithms to predict traffic patterns and adjust routes dynamically. Additionally, it may integrate with navigation systems to provide turn-by-turn directions and real-time updates to the delivery driver. The goal is to minimize travel time, reduce fuel consumption, and enhance overall delivery performance. The system may also include features for tracking delivery status, managing exceptions, and optimizing multiple deliveries in a single trip. By leveraging advanced routing algorithms and real-time data, the system aims to improve the efficiency and reliability of delivery operations.
11. The non-transitory computer readable medium of claim 8, wherein the current interaction is a natural language interaction.
12. The non-transitory computer readable medium of claim 8, wherein the image concept vector for the image was determined by: determining one or more concepts associated with the image; determining a score for the one or more concepts associated with the image and applying the score for the one or more concepts associated with the image to a concept net relationship matrix.
13. The non-transitory computer readable medium of claim 8, wherein the canonical concept is based on one or more previous user interactions in a session associated with the user.
14. The non-transitory computer readable medium of claim 8, wherein the image is selected based on its similarity to other images in the set of images.
This invention relates to image selection techniques in computer vision systems. The problem addressed is efficiently selecting images from a set based on similarity to other images in the set, which is useful for applications like image clustering, recommendation systems, or dataset curation. The system involves a computer-readable medium containing instructions for processing a set of images. The instructions include comparing each image in the set to others using a similarity metric, such as visual feature matching or deep learning-based embeddings. Images are then ranked or filtered based on their similarity scores. The selection process may involve clustering similar images, identifying outliers, or prioritizing images that are representative of the set. The similarity comparison may use pre-trained neural networks to extract high-dimensional feature vectors from each image, which are then compared using distance metrics like cosine similarity or Euclidean distance. The system can also incorporate metadata or user preferences to refine the selection. The output is a subset of images that are most similar to others in the set, which can be used for further analysis or presentation. This approach improves efficiency by reducing the number of images processed in downstream tasks while maintaining relevance. It is particularly useful in large-scale image datasets where manual selection is impractical. The method can be applied in various domains, including medical imaging, surveillance, and e-commerce product recommendations.
16. The method of claim 15, wherein determining the destination associated with the canonical concept is determined comprises: determining a score for the canonical concept for each destination in the set of destinations, wherein determining a score for the canonical concept for the destination comprises accessing the destination concept vector associated with the destination to determine the score for the canonical concept for the destination and selecting the destination based on the score for the canonical concept for each destination in the set of destinations.
This invention relates to a method for determining a destination associated with a canonical concept in a system that processes and routes information. The problem addressed is efficiently selecting an appropriate destination from a set of possible destinations based on semantic similarity between a canonical concept and destination-specific concept vectors. The method involves determining a score for the canonical concept relative to each destination in the set. To compute this score, the system accesses a destination concept vector associated with each destination. The destination concept vector represents the semantic characteristics of the destination. The score reflects the degree of similarity or relevance between the canonical concept and the destination's concept vector. After scoring all destinations, the system selects the destination with the highest score, indicating the most semantically relevant or appropriate destination for the canonical concept. This approach leverages vector-based semantic analysis to improve routing accuracy, ensuring that the canonical concept is directed to the most suitable destination based on contextual relevance. The method is particularly useful in systems requiring intelligent information routing, such as search engines, recommendation systems, or automated content distribution platforms.
17. The method of claim 15, wherein the destination is a neighborhood, apartment, or house.
This invention relates to a method for optimizing delivery routes, particularly for last-mile logistics in urban and residential areas. The problem addressed is the inefficiency of traditional delivery methods, which often result in excessive travel time, fuel consumption, and delays due to unpredictable factors like traffic, parking, and access restrictions in densely populated or complex residential zones. The method involves dynamically determining an optimal delivery route based on real-time data, including traffic conditions, delivery vehicle capacity, and recipient availability. It incorporates machine learning algorithms to predict the most efficient path, considering factors such as neighborhood layout, apartment building access, and house-specific constraints. The system may also integrate with smart home devices or recipient schedules to ensure timely and secure deliveries. A key aspect of the method is its ability to adapt to different destination types, including neighborhoods, apartments, and individual houses. For apartment deliveries, the system may prioritize buildings with available parking or elevators, while for houses, it may optimize for proximity and ease of access. The method also accounts for multi-unit deliveries, such as apartment complexes, by grouping deliveries to minimize redundant travel. By leveraging real-time data and predictive analytics, the method reduces delivery times, lowers operational costs, and improves overall logistics efficiency in residential and urban environments.
18. The method of claim 15, wherein the current interaction is a natural language interaction.
19. The method of claim 15, wherein the image concept vector for the image was determined by: determining one or more concepts associated with the image; determining a score for the one or more concepts associated with the image and applying the score for the one or more concepts associated with the image to a concept net relationship matrix.
20. The method of claim 15, wherein the canonical concept is based on one or more previous user interactions in a session associated with the user.
21. The method of claim 15, wherein the image is selected based on its similarity to other images in the set of images.
This invention relates to image selection techniques, specifically for choosing an image from a set of images based on its similarity to other images in the set. The method addresses the challenge of efficiently selecting representative or relevant images from a collection, which is useful in applications like image retrieval, clustering, or recommendation systems. The core process involves analyzing the visual or semantic features of images to determine their similarity to one another. By comparing these features, the method identifies and selects an image that closely matches the characteristics of other images in the set. This ensures that the chosen image is representative of the group, improving accuracy in tasks such as content-based filtering or automated curation. The technique may involve computational models like neural networks or traditional similarity metrics to assess image relationships. The selection process can be applied iteratively or in real-time, depending on the application requirements. This approach enhances efficiency in image processing workflows by reducing redundancy and improving relevance in image selection tasks.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
November 13, 2020
October 25, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.